Intelligent generation of job profiles

Abstract

In one aspect, an example methodology implementing the disclosed techniques includes, by a computing device, receiving activity data corresponding to an activity of an employee and deriving context and information from the activity data. The method also includes, by the computing device, associating the context and information to a job position and storing the context and information within a repository. The method further includes, by the computing device, responsive to a request to generate a job profile for the job position, retrieving, from the repository, the context and information associated with the job position and generating the job profile based on the retrieved context and information.

What is claimed is: 
     
         1 . A computer-implemented method for generating job profile text using a trained neural network, the method comprising:
 building, by a computing device, a machine learning (ML) model by:
 generating a modeling dataset from a text corpus and a sentiment corpus using a natural language process, wherein the modeling dataset comprises a plurality of training samples, each training sample including feature values derived from the text corpus and the sentiment corpus and a corresponding job-profile text sequence; 
 implementing through a recurrent neural network, by the computing device, a unidirectional long short-term memory model; and 
 training the unidirectional long short-term memory model to generate job profile text using a plurality of training samples, the plurality of training samples generated from the modeling dataset; 
   receiving, by a computing device, activity data corresponding to activities of an employee over a period of time;   deriving from the activity data, by the computing device, context and information by organizing the activity data into an activity graph and generating, from the activity graph, the context and information, the context and information comprising an intent derived using a first bidirectional long short-term memory model and a sentiment derived using a second bidirectional long short-term memory model, wherein the first and second bidirectional long short-term memory models process sequences of tokens in both forward and backward directions using the activity graph to derive the intent and the sentiment, thereby improving a contextual representation of the activity data used by the machine learning model;   associating, by the computing device, the context and information to a job position;   storing, by the computing device, the context and information within a personal knowledge repository; and   responsive to a request to generate a job profile for the job position:
 retrieving, by the computing device from the repository, the context and information associated with the job position; 
 inputting, by the computing device, the context and information to the ML model; and 
   generating, from the ML model, by the computing device, a job profile text.   
     
     
         2 . The method of  claim 1 , wherein the activities includes a written communication with a colleague. 
     
     
         3 . The method of  claim 1 , wherein the activities includes an oral communication with a colleague. 
     
     
         4 . The method of  claim 1 , wherein the activities includes an access of a document related to the job position. 
     
     
         5 . The method of  claim 1 , wherein the activities includes a surfing task. 
     
     
         6 . The method of  claim 1 , wherein the context and information include one or more of a role, a domain, an intent, and a sentiment. 
     
     
         7 . The method of  claim 1 , wherein the activity data is received over a secure communication channel. 
     
     
         8 . The method of  claim 1 , wherein generating the job profile text includes completing a job profile template generated for the job position. 
     
     
         9 . The method of  claim 1 , wherein the context and information are stored in an encrypted format within the repository. 
     
     
         10 . A computing device comprising:
 one or more non-transitory machine-readable mediums configured to store instructions; and   one or more processors configured to execute the instructions stored on the one or more non-transitory machine-readable mediums, wherein execution of the instructions causes the one or more processors to carry out a process comprising:
 building a machine learning (ML) model by:
 generating a modeling dataset from a text corpus and a sentiment corpus using a natural language process, wherein the modeling dataset comprises a plurality of training samples, each training sample including feature values derived from the text corpus and the sentiment corpus and a corresponding job-profile text sequence; 
 implementing through a recurrent neural network a unidirectional long short-term memory model; and 
 training the unidirectional long short-term memory model to generate job profile text using a plurality of training samples, the plurality of training samples generated from the modeling dataset; 
 
 receiving activity data corresponding to activities of an employee over a period of time; 
 deriving from the activity data context and information by organizing the activity data into an activity graph and generating, from the activity graph, the context and information, the context and information comprising an intent derived using a first bidirectional long short-term memory model and a sentiment derived using a second bidirectional long short-term memory model, wherein the first and second bidirectional long short-term memory models process sequences of tokens in both forward and backward directions using the activity graph to derive the intent and the sentiment, thereby improving a contextual representation of the activity data used by the machine learning model; 
 associating the context and information to a job position; 
 storing the context and information within a personal knowledge repository; and 
 responsive to a request to generate a job profile for the job position:
 retrieving, from the repository, the context and information associated with the job position; 
 inputting the context and information to the ML model-to-generate; and 
 generating from the ML model a job profile text. 
 
   
     
     
         11 . The computing device of  claim 10 , wherein the activities includes one of a written communication, an oral communication, an access of a document related to the job position, and a surfing task. 
     
     
         12 . The computing device of  claim 10 , wherein the context and information include one or more of a role, a domain, an intent, and a sentiment. 
     
     
         13 . The computing device of  claim 10 , wherein the activity data is received over a secure communication channel. 
     
     
         14 . The computing device of  claim 10 , wherein generating the job profile includes completing a job profile template generated for the job position. 
     
     
         15 . A computer-implemented method for generating job-profile text using a trained neural network, the method comprising:
 building, by a computing device, a machine-learning (ML) model implemented as a recurrent neural network comprising a unidirectional long short-term memory (LSTM) layer and configured to generate job-profile text, the building comprising:
 generating, using natural language processing, a modeling dataset from a text corpus and a sentiment corpus; and 
 training the recurrent neural network using training samples from the modeling dataset; 
   receiving, by the computing device, activity data corresponding to activities of an employee over a period of time;   deriving, by the computing device, context and information from the activity data, the deriving comprising organizing the activity data into an activity graph and generating, from the activity graph, feature values including an intent value derived using a first bidirectional LSTM model and a sentiment value derived using a second bidirectional LSTM model, wherein the first and second bidirectional long short-term memory models process sequences of tokens in both forward and backward directions using the activity graph to derive the intent and the sentiment, thereby improving a contextual representation of the activity data used by the machine learning model;   associating, by the computing device, the context and information with a job position;   storing, by the computing device, the context and information in a personal knowledge repository implemented using non-transitory computer-readable storage and maintained in association with the job position; and   in response to receiving, by the computing device, a request to generate a job profile for the job position:
 retrieving, from the personal knowledge repository, the context and information associated with the job position; 
 providing the context and information as input to the trained recurrent neural network; and 
 generating, by the trained recurrent neural network, job-profile text for the job position based on the context and information. 
   
     
     
         16 . The machine-readable medium of  claim 15 , wherein the activities includes one of a written communication, an oral communication, an access of a document related to the job position, and a surfing task.